Saturday, March 28, 2015

Team Pace and Efficiency by Possession Type

As a follow up to last week's post on offensive and defensive pace, here is a deeper dive into the offensive side of things. In that original post, I found that the Golden State Warriors have the league's third slowest pace on defense, despite leading the NBA in overall pace. The Warriors manage this by being really, really, really, ridiculously fast on offense, averaging nearly a second less per possession than the next fastest offensive team, the Celtics. When charted, their outlier status is clear.

Offensive and defensive pace were calculated by analyzing the detailed play by play data for each game (the same data that powers my win probability graphs and box scores). Using that same data set, we can go even further, classifying possesions by how they begin.

The vast majority of possessions start one of three ways: after an opponent made shot, after a defensive rebound, or after an opponent turnover. The table below summarizes average pace and efficiency (points per possession) for those three scenarios:

Sunday, March 22, 2015

ESPN's Tom Haberstroh Makes the Case for Anthony Davis as MVP

At ESPN Insider, Tom Haberstroh uses win probability added to argue the case for Anthony Davis as league MVP. From a win probability perspective, it's a very compelling case, as I have pointed out before. The Brow's box score stats are gaudy, but still undersell his contributions to the Pelicans when it mattered most.

In addition to being a good read in general, Tom provides a good overview of what win probability added means, and what it does (and does not) count. From the article (sorry, it's behind a paywall):

At any given point in a game, each team has a win probability depending on the game state (quarter, clock, margin and possession). If a player makes a basket, he improves the team's win probability. If he misses the shot or turns the ball over, he decreases their odds of winning. Intuitive stuff, right? Michael Beuoy, who runs Inpredictable.com, has written an algorithm that adds up all the credit and debits that a player accumulates during a game to arrive at a summed total of win probability added (WPA). Beuoy then takes a step further and aggregates all of those game totals for the entire season.
To be clear, player WPA only looks at the basic stuff: field goal attempts (made or missed), free throws (made or missed) and turnovers. After that, it adjusts for the game state because the gravity of the situation matters. For example, if a player misses a potential game-tying free throw at the end of the game, it's a much more devastating blow than if he misses a free throw in the opening minute of the game. By pinning it to a game state, WPA offers context to the box score numbers.

Offensive versus Defensive Pace in the NBA

Last night's game between the Golden State Warriors and the Utah Jazz was billed by some as a matchup between the league's best offense (the Warriors) and the league's best defense (the Jazz). And while that is technically true when measured by points per game, the statistically savvy know this is a misleading way of judging team offense and defense.

The Utah Jazz have the slowest pace in the league, averaging just 90.1 possessions per game. This leads to fewer scoring opportunities on both offense and defense. At the opposite end of the spectrum are the Warriors, who lead the league with an average of 98.3 possessions per game. If we adjust for points scored per possession, it turns out that the best defense in the league did take the floor last night - but it was Golden State, not Utah. The Utah Jazz allow a merely average 1.05 points per possession, while the Warriors lead the league with an average 1.01 points allowed for each possession.

Wednesday, March 18, 2015

Relatively speaking, the odds of a perfect bracket are pretty good this year

Last year, I attempted to calculate the odds of filling out a perfect bracket in the NCAA Tournament. Using my betting market rankings, I found that the odds of a perfect bracket are about 50 billion to 1 (give or take 20 billion). This year, however, if I use the same methodology, the perfect bracket odds are "just" 6 billion to 1. Nate Silver at FiveThirtyEight is even more optimistic. Using their model, they estimate a perfect bracket to be a 1.6 billion to 1 proposition.

In case you're looking for help with your bracket picks this year, here is a link to that 6 billion to 1 bracket. Of the four regions, the West has the best chance of coming in perfectly. Here are the odds for each region, plus the final four:

• Midwest: 270 to 1
• West: 148 to 1
• East: 167 to 1
• South: 203 to 1
• Final Four: 4 to 1

If you follow the link, you'll find that this bracket has quite a lot of chalk. There are very few upsets predicted, especially as we get to the later rounds. Because this bracket adheres to the general consensus so closely, it is probably not the best one to use for an office pool, where you'll need to get lucky with a few upsets in order to come out on top.

But all upsets are not created equal, and we can use my rankings to find the most likely upsets to pick. I used the following rule to create a more upset-heavy bracket: Pick the lower seed in a matchup provided they were no worse than a projected two point underdog to their opponent. See here for the results. There's still not a lot of upsets here, but we do have 11 seed Texas surviving all the way to the Elite Eight, and two seed Virginia squaring off against Kentucky in the final.

Monday, March 16, 2015

Do Teams Shoot Threes More Often When Down By Three?

The nice thing about building your own NBA database is the ability to satisfy your idle curiosity as a fan (the drawback: creeping madness ensuing from sifting through hundreds, if not thousands of lines of code for an errant parentheses).

When watching games, it always seemed to me that teams were more likely to attempt three point shots when trailing by three. Of course, that is rational and to be expected as the game nears its final minute (expected, but not easy). But this bias for three pointers seems to occur throughout the game. When there is plenty of time left, one would presume that teams should be taking the best shot available - one which maximizes expected points, regardless of whether it is a two or a three. But the psychological pull of "evening the score" seems to bias shot selection when trailing by three.

Monday, March 9, 2015

Clutch Shooting Isn't Easy (unless it's a free throw)

Last Thursday night, the Bulls trailed the Thunder at home by 1 point with 4.9 seconds left on the clock. The Bulls' inbounds play was designed to get a shot for Pau Gasol, but Gasol improvised instead with a touch pass to E'Twaun Moore on the perimeter. Moore's three point shot was good and gave the Bulls a 2 point lead with just two seconds to go (cue Benny the Bull).

You don't need any fancy analytics to know that Moore's shot was important. But we can use win probability to quantify just how important that shot was. Moore's three pointer gave the Bulls a win probability of 92.9%. If he had missed, their win probability would have been just 6.4%, a swing of 86.5% between the two outcomes (note: you can find these numbers for any shot by mousing over the leverage bar charts at the bottom of each win probability graph).

Sunday, March 1, 2015

The Trailblazers have had the most heartbreaking losses this season (on average)

 Choking away victory isalive in Portland
I have a new NBA feature on the site: Team Season Profiles. This page aggregates win probability information at the season and team level. My previous post covered the four factors numbers on the right side of the table. This post will cover the columns on the left labelled "Game Profile".

The first column, "EI", is the average Excitement Index for each team's games. This is a "live" version of the table I shared in my recent FiveThirtyEight article on "exciting" NBA games. The Excitement Index is measured at a game level, and represents how much the win probability graph "travelled" over the course of the game. The Phoenix Suns still lead the league in excitement, with an average index of 6.78, but the Lakers have now overtaken the Spurs at the number two spot.

There are also two columns called "wCB" and "lCB". This is the average comeback factor for each team, split by games won versus games lost. The comeback factor is the winning team's odds of winning at their lowest point in the game. The higher the comeback factor, the bigger the comeback.

When sorted by losing comeback factor (lCB), the Portland Trailblazers "lead" the league, meaning the games they lost have a high average comeback factor, such as allowing the Mavericks to comeback from 830-1 odds last month. You could also consider this a "choke" index. Mathematical note: to minimize the effect of outliers, I take the geometric mean of the comeback factor, rather than a standard arithmetic mean.

One of the nice side effects of the NBA's data explosion is that no matter how awful your team is, you're bound to find something at which they excel. The Philadelphia 76ers don't win too often, but when they do, it's dramatic, as they have the highest average comeback factor for their wins (all 13 of 'em).

Explaining Team Win-Loss Records

A useful, sometimes ignored distinction in sports analytics is the difference between a "narrative" stat and a "predictive" stat. Narrative stats tell you what happened. Predictive stats tell you what will happen. Predictive stats are what your general managers, bookies, and gamblers care about, but it's usually the narrative stats that get the most attention. In addition, narrative stats are often mistaken for their predictive cousins (it happens with elections too).

Win probability added is, perhaps, the ultimate narrative stat. It can distill any play down to its impact on what we care about the most: winning. In last week's post on how NBA games are won, I used my win probability model to assess the relative importance of the "four factors" of basketball success, leading to the ground breaking conclusion that shooting is very, very important to the game of basketball (take that, Chuck).

In this post, I introduce a new feature to the site: NBA Team Profiles. In the same way I can use my narrative win probability added stat to take stock of the four factors, I can do the same for team win-loss records. The new feature breaks down a team's win loss record according to the four factors, and further split by offense and defense. Here is an example: